How to solve unknown elements of a matrix in mle function?

3 次查看（过去 30 天）

Mithun Ghosh 2018-9-26

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/420909-how-to-solve-unknown-elements-of-a-matrix-in-mle-function

评论： Viren Gupta 2018-10-4

I have been trying to find the unknown value of a basis function by using mle estimation method. My basis function consists of basis(x,k) where x is the data and k is the unknown parameter with (5X1) dimension say[k(1), k(2),k(3),k(4),k(5)]. My liklihood function estimation is like:

function Z=bbasis(x,k)
k=[k(1), k(2),k(3),k(4),k(5)];
z=[];
  for i=1:5
       Z=[x-k(i)];
       z=[z,Z];
  end
end
x=rand(20,1);
y=rand(20,1);
F=ones(20,1); 
f = @(x,w) (2*3.1416)^-(20)*(det(bbasis(x,w)*rand(10,10)*bbasis(x,w)'))^(-.5)*exp(-.5*(y-F)'*inv(bbasis(x,w)*rand(10,10)*bbasis(x,w)')*(y-F))
lk=mle(x,'pdf','f','start',[0 0 0 0 0 ])
*I am getting the following error:* 
Error using mlecustom (line 93)
The 'pdf' parameter value must be a function handle or a cell array
containing a function handle.
Error in mle (line 237)
            phat = mlecustom(data,varargin{:});

3 个评论
显示 1更早的评论隐藏 1更早的评论

Mithun Ghosh 2018-10-1

Thanks, I corrected that dimensional mismatch problem and redefine pdf which should be okay. But it is still giving me the same error.

Viren Gupta 2018-10-4

have you updated the changes in the question above? If yes, then still there is dimension problem. bbasis1 returns 20*1 matrix and multiplying that with rand(10,10) is wrong. I suggest you to try giving some sample inputs to your pdf function and see if that works before working on mle first.

请先登录，再进行评论。

请先登录，再回答此问题。